Learning-by-Exporting under Credit Constraints1

Robert Petrunia (Lakehead University)

joint with Kim Huynh (Bank of Canada),Joel Rodrigue (Vanderbilt U), Walter Steingress (Bank of Canada)

2019 Canadian Stata ConferenceMay 30, 2019 - Banff

1The views expressed in this paper are those of the authors and do not necessarilyreflect those of the Bank of Canada.

1/15

Main question

Improved access to foreign markets increases demand and encouragesfirms to invest.

Financial constraints may prevent firms to exploit these opportunities.

2/15

Contribution and findings

Built framework to motivate firms’ export and investment decisions

Relationship: return on exporting and firm’s access to credit markets

Firms’ financial constraints are unobserved

Use marginal treatment framework to quantify selection effect

Results

Exporters have higher productivity

Exporters have lower debt to asset ratios

Firms that are more likely to be induced to export acquire moredebt

→ positive selection is suggestive of financial constraints

3/15

Literature

Learning from exporting

Clerides, Lach and Tybout, (1996), Bernard and Jensen (1999), Baldwinand Gu (2003)

Aw, Roberts and Winston (2007), De Loecker (2007)

Lileeva and Trefler (2010), Aw, Roberts and Xu (2011)

Credit constraints and exporting

Greenaway et al. (2007), Manova et al. (2009), Minetti and Zhu (2011),Amiti and Weinstein (2011), Manova (2013)

Caggese and Cunat (2013), Brooks and Dovis (2011), Leibovici (2014),Kohn et al. (2015)

Marginal treatment framework

Heckman and Vytlacil (2005), Carneiro, Heckman, Vytlacil (2010)

4/15

Theoretical framework

Heterogeneous firms model of international trade

Firms can invest in productivity enhancing technology and\or exporting

Firms can borrow from investors and pledge tangible assets as collateral

Resulting constraints in Model:

1 Incentive compatibility const. (IC) → E[return invest] ≥ E[return noinvest]

2 Banks’ participation const. (PC) → E[return invest] ≥ bank loan3 Export constraint (EC) → Need to finance fixed cost to export

5/15

Theoretical framework

Heterogeneous firms model of international trade

Firms can invest in productivity enhancing technology and\or exporting

Firms can borrow from investors and pledge tangible assets as collateral

Resulting constraints in Model:

1 Incentive compatibility const. (IC) → E[return invest] ≥ E[return noinvest]

2 Banks’ participation const. (PC) → E[return invest] ≥ bank loan

3 Export constraint (EC) → Need to finance fixed cost to export

5/15

Theoretical framework

Heterogeneous firms model of international trade

Firms can invest in productivity enhancing technology and\or exporting

Firms can borrow from investors and pledge tangible assets as collateral

Resulting constraints in Model:

1 Incentive compatibility const. (IC) → E[return invest] ≥ E[return noinvest]

2 Banks’ participation const. (PC) → E[return invest] ≥ bank loan3 Export constraint (EC) → Need to finance fixed cost to export

5/15

Theoretical framework

Heterogeneous firms model of international trade

Firms can invest in productivity enhancing technology and\or exporting

Firms can borrow from investors and pledge tangible assets as collateral

Resulting constraints in Model:

1 Incentive compatibility const. (IC) → E[return invest] ≥ E[return noinvest]

2 Banks’ participation const. (PC) → E[return invest] ≥ bank loan

3 Export constraint (EC) → Need to finance fixed cost to export

5/15

Theoretical framework

Heterogeneous firms model of international trade

Firms can invest in productivity enhancing technology and\or exporting

Firms can borrow from investors and pledge tangible assets as collateral

Resulting constraints in Model:

1 Incentive compatibility const. (IC) → E[return invest] ≥ E[return noinvest]

2 Banks’ participation const. (PC) → E[return invest] ≥ bank loan3 Export constraint (EC) → Need to finance fixed cost to export

5/15

Decision to invest and exportMarginal returns

Productivity (φ0)

Return (φ1 − φ0)

IC

Firms with higher returns will choose to export and invest.

Conditional on initial productivity and financial conditions.

6/15

Decision to invest and exportMarginal returns

PC

Productivity (φ0)

Return (φ1 − φ0)

IC

invest

credit

not-investing

constraint

Firms with higher returns will choose to export and invest.

Conditional on initial productivity and financial conditions.

6/15

Decision to invest and exportMarginal returns

PC

Productivity (φ0)

Return (φ1 − φ0)

IC

invest+export

credit

not-investing

EC

not-exporting

export onlyconstraint

Firms with higher returns will choose to export and invest.

Conditional on initial productivity and financial conditions.

6/15

Decision to invest and exportMarginal returns

PC

Productivity (φ0)

Return (φ1 − φ0)

IC

invest+export

credit

not-investing

EC

not-exporting

export onlyconstraint

Firms with higher returns will choose to export and invest.

Conditional on initial productivity and financial conditions.6/15

Going to empirics

Main identification issues:

credit constraints are not observable

Our solution:

Estimate marginal returns to exporting

Firms with higher returns will choose to export and acquire more debt

→ positive selection

7/15

Data - ASM/T2

Two Sources linked:

ASM - Annual Survey of Manufacturers

T2 corporate tax records

ASM-T2: ASM (Plant Level) linked with T2 (Firm Level)

Annual Data: 2000-2010

Manufacturers

Firm-level variables are common to all plants of the firm.

8/15

Data - ASM/T2

ASM production/export variables:

Value Added, Employment (production and non-production), Salary andWages,

Sales, Material and Supplies Costs, Fuel and Electricity Costs, Value ofShipments,

Value of Shipments Exported, NAICS classification, Plant Age

T2 corporate balance sheet variables

Assets, Tangible Assets, Sales, Profits, Equity,

Total Debt, Total Long-term Liabilities, Working Capital, CorporationType,

Firm corporate start year

9/15

Estimation equation

Reg.: exporters (treated j = 1) and non-exporters (untreated j = 0)

Y[j ],it = β[j ]X[j ],it +K[j ](p) + eit (1)

Y : leverage ratio of firm i in year t (proxy for access to credit)

X : initial leverage ratio, value added labor productivity, sales, age,industry dummies

K control function: 3rd order polynomial

Andresen (2018) Stata Journal - MTEFE module

Instruments:

Industry-specific US-CA Real Exchange Rate in year t

Changes in US tariffs after China’s entry to the WTO

10/15

ResultsDep. variable Leverage ratio

untreated treated vs. untreated

Init. leverage ratio 0.8367*** 0.0922***(0.0131) (0.0302)

Init. labor prod -0.1960*** 0.1900***(0.0276) (0.0610)

Age -0.0274 0.0756***(0.0151) (0.0329)

Age squared -0.0241*** 0.0444***(0.0046) (0.0099)

Number of obs 415,773 415,773Replications 100 100

Initial financial conditions are important

Initially less productive firms have higher leverage ratio

Exporters: higher leverage ratios, more productive and older.

11/15

ResultsDep. variable Leverage ratio

untreated treated vs. untreated

Init. leverage ratio 0.8367*** 0.0922***(0.0131) (0.0302)

Init. labor prod -0.1960*** 0.1900***(0.0276) (0.0610)

Age -0.0274 0.0756***(0.0151) (0.0329)

Age squared -0.0241*** 0.0444***(0.0046) (0.0099)

Number of obs 415,773 415,773Replications 100 100

Initial financial conditions are important

Initially less productive firms have higher leverage ratio

Exporters: higher leverage ratios, more productive and older.

11/15

Treatment effects

(1)

ATE 0.850***(0.061)

ATT 1.424***(0.105)

ATUT 0.512***(0.111)

LATE 0.649***(0.043)

Test of observable heterogeneity, p-value 0.0000Test of essential heterogeneity, p-value 0.0000

Exporting increases the leverage ratio.

ATT>ATE>ATUT ⇒ positive selectionfirms with higher expected returns acquire more debt

→ consistent with presence of financial constraints12/15

Thanks/Merci

13/15

Appendix

13/15

Summary statistics

Non-exporters Exporters Difference

Mean Std. dev. Mean Std. dev. t-stat

Assets (thous.) 7006.3 26274.1 18535.6 40003.3 -123.0

Debt (thous.) 3822.5 13715.4 9867.4 20665.1 -124.3

Sales (thous.) 7302.0 24274.7 19276.7 37024.3 -138.1

Employment 15.31 24.15 38.3 44.7 -235.1

Profit (thous.) 1633.4 4888.1 3923.5 7134.9 -134.7

Value added labor prod (thous.) 77.2 42.1 89.3 47.8 -94.0

Debt to asset ratio 0.794 0.556 0.704 0.451 60.4

Age 9.83 5.526 10.19 5.71 -22.4

Observations 298890 201369

Exporters are larger, older and have higher productivity

Lower debt to asset ratio

13/15

ResultsDep. variable Labour productivity

untreated treated vs. untreated

Init. leverage ratio -0.0024*** -0.0003(0.002) (0.001)

Init. labor prod 0.108*** 0.421***(0.0138) (0.0201)

Age 0.193 -0.335***(0.0132) (0.021)

Age squared -0.002*** 0.0004***(0.00003) (0.00005)

Number of obs 415,773 415,773Replications 100 100

More productive firms have lower initial debt to asset ratio

Initially more productive firms remain more productive

Exporters: more productive and younger.

14/15

ResultsDep. variable Labour productivity

untreated treated vs. untreated

Init. leverage ratio -0.0024*** -0.0003(0.002) (0.001)

Init. labor prod 0.108*** 0.421***(0.0138) (0.0201)

Age 0.193 -0.335***(0.0132) (0.021)

Age squared -0.002*** 0.0004***(0.00003) (0.00005)

Number of obs 415,773 415,773Replications 100 100

More productive firms have lower initial debt to asset ratio

Initially more productive firms remain more productive

Exporters: more productive and younger.

14/15

Treatment effects

(1)

ATE 0.23*(0.135)

ATT 2.203***(0.329)

ATUT -1.003***(0.205)

LATE 1.443***(0.246)

Test of observable heterogeneity, p-value 0.0000Test of essential heterogeneity, p-value 0.0000

Exporting increases productivity.

ATT>ATE>ATUT ⇒ positive selectionFirms that are more likely to choose to export become more productive.

→ firms with higher expected returns expand productivity more.15/15